1. 2nd Stage Development of an Autonomous Search and Rescue UAV
Team 12: Matthias Clarke, Devin Justice, Trent Loboda, Cody Rochford, Marcus Yarber, Qinggele ‘Gale’ Yu
Presenters: Matthias Clarke, Marcus Yarber, Qinggele ‘Gale’ Yu
Advisor: Dr. Alvi 1

2. Project Introduction Previous Work Current work conclusion
Student Unmanned Aerial Systems (SUAS) Competition
Simulated search and rescue of a hiker
Tasks:
Autonomous flight
Waypoint capturing, navigation, take off, and landing
Payload delivery
Target detection
Object avoidance
Figure 1. Mission map for the SUAS 2017 Competition.
Matthias Clarke
Project Introduction Previous Work Current work conclusion 2

3. Project Introduction Previous Work Current work conclusion
Development Stages
1st stage development:
Design, build, test UAV system
2nd stage development:
Upgrade electronics package
Implement:
Target color detection
Target shape detection
Target alphanumeric detection
Dynamic target detection
Payload delivery
Object avoidance
6’-7”
Figure 2. Current UAV developed by team 8 in 2016.
Matthias Clarke
Project Introduction Previous Work Current work conclusion 3

4. Project Introduction Previous Work Current work conclusion
Objectives
Objectives:
Select and develop electronics payload.
Design, build, and integrate payload delivery mechanism.
Select and integrate lightweight landing gear configuration.
Create programs to detect and classify stationary targets.
Create programs to detect and classify emerging targets.
Develop control system to autonomously avoid objects. 3
Figure 3. Stationary target example.
Matthias Clarke
Project Introduction Previous Work Current work conclusion 4

5. Project Introduction Previous Work Current work conclusion
Constraints
Constraints:
Maximum takeoff weight must be less than 55 lbs.
Maximum speed must be less than 70 Knots Indicated Air Speed (KIAS).
Must be capable of operation with 15 knots winds with gusts up to 20 knots.
Must be capable of operation in 110°F and average over 100°F for over 12 hours.
Solutions:
Wind load calculations have been conducted.
Finite element analysis to be done on heat transfer to electronics from environment.
Matthias Clarke
Project Introduction Previous Work Current work conclusion 5

6. Payload Delivery Mechanism
Figure 4. 3D Model Rendering of Latch-Servo Design.
Figure 5. 3D Animation of Latch-Servo Design.
Marcus Yarber
Project Introduction Previous Work Current work conclusion 6

7. Payload Delivery Mechanism
Figure 6. First Iteration Prototype.
Marcus Yarber
Project Introduction Previous Work Current work conclusion 7

8. Electrical Systems Design
System Requirement Solution
Autonomy Autopilot component
Navigation GNSS/IMU
Target detection Camera
Communication Multiple antennas
Image Processing CPU
Marcus Yarber
Project Introduction Previous Work Current work conclusion 8

9. Zubax Global Navigation Satellite System (GNSS)
Ordered Electronics
Table 1. Ordered electronics and cost.
Electrical Component
Cost (Dollars)
Odroid C2 Motherboard 40
Zubax Global Navigation Satellite System (GNSS)
129
Sony Camera 90
Wi-Fi Antennas 26
Total:
285
Marcus Yarber
Project Introduction Previous Work Current work conclusion 9

10. Project Introduction Previous Work Current work conclusion
Electronics Update
Figure 7. Original top-level electronics design.
Marcus Yarber
Project Introduction Previous Work Current work conclusion 10

11. Project Introduction Previous Work Current work conclusion
Electronics Update
Figure 8. Proposed top-level electronics design.
Marcus Yarber
Project Introduction Previous Work Current work conclusion 11

12. Landing Gear Selection
Problem:
Current landing gear risks damaging rotors upon landing.
Solution:
New landing gear under development, allows clearance for rotors.
New landing gear will have a rear castor wheel which will increase stability during takeoff and landing.
Figure 9. Forward strut example of new landing gear.
Marcus Yarber
Project Introduction Previous Work Current work conclusion 12

13. Design of Experiments: Target Detection
Taguchi method 2 level fractional factorial design used to design experiments for color detection and alphanumeric symbol detection.
Variables:
Brightness High/Low
Target size Large/Small
Contrast High/Low
Distance Far/Close
Motion Dynamic/Static
𝑛= 2 𝑘
𝑛≜𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑒𝑥𝑝𝑒𝑟𝑖𝑚𝑒𝑛𝑡𝑠
𝑘 ≜𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟𝑠
Preliminary testing: 32 trials
Further testing of color, shapes, and symbols will make use of Taguchi orthogonal arrays.
Gale Yu
Project Introduction Previous Work Current work conclusion 13

14. Design of Experiments: Flight
Order of testing:
Motors
Pixhawk
GPS
Sensors
Waypoint capture
Path following
Take-off/Landing
Object avoidance
Fall Semester
Spring Semester
Gale Yu
Project Introduction Previous Work Current work conclusion 14

15. Object Avoidance: Preliminary Approach
Paths:
Normal (Waypoints, etc.)
Avoidance (Avoid Obstacle)
Return (Return To Normal)
Ranges:
Collision (Closest Range)
Threat (Immediate Action)
Invader (Detectable/In UAV Flying Area)
PDF: Moving Obstacle Avoidance for Unmanned Aerial Vehicles by Yucong Lin
Focuses Mainly on Obstacle Avoidance (Not The Return Path,etc.)
Attractive Forces Include: Normal Path/ Target or Goal
Repulsive Forces Include: Geo-fence, The Obstacle, Ground/Max Alt
Avoidance Path Computation:
Min/Max Climb Angles (Pitch)
Min/Max Steering Angles (Yaw)
Min/Max Speed & Acceleration
Attractive & Repulsive Forces
Figure 10. Object avoidance working principle.
Gale Yu
Project Introduction Previous Work Current work conclusion 15

16. Dynamic Target Detection: Approach
Compare current image with previous; use differences to detect changes.
Challenge: constantly changing environment.
Proposed solution: if large changes are found, set image as new background.
Figure 11. Motion detection progression.
Gale Yu
Project Introduction Previous Work Current work conclusion 16

17. Project Introduction Previous Work Current work conclusion
Future Work
Test code for color and alphanumeric detection.
Test electronics upon delivery and test flight capabilities.
Integrate payload delivery mechanism.
Continue work on first iteration of object avoidance code.
Continue work on first iteration of dynamic target detection
Figure 12. Letter recognition processing.
Gale Yu
Project Introduction Previous Work Current work conclusion 17

18. Project Introduction Previous Work Current work conclusion
Challenges
Target detection:
Recognize alphanumeric symbols and shapes, reliant on camera capability and distance.
Emergent target detection in a constantly changing environment is less prevalent in open source code.
Recognize objects while flying with the propellers in the horizontal position.
Development of transitional propeller control system.
Object avoidance:
Develop evasive maneuvering within the limited flight envelope for flying wing design.
Write a robust program capable implementation with hardware for realistic application.
Gale Yu
Project Introduction Previous Work Current work conclusion 18

19. Project Introduction Previous Work Current work conclusion
Electronics have been ordered.
Payload delivery mechanism has been 3D printed.
Testing of color detection and alphanumeric detection has been outlined.
Test flight tasks have been determined.
Object avoidance and dynamic target detection approaches have been outlined.
Gale Yu
Project Introduction Previous Work Current work conclusion 19

20. Project Introduction Previous Work Current work conclusion
References
Competition Rules SUAS (2017).
K. Aley, J. Denman, D. Fitzpatrick, C. Mard, P. McGlynn, and K. Ijagbemi, "Needs Assessment" Sep. 25, 2015.
K. Aley, J. Denman, D. Fitzpatrick, C. Mard, P. McGlynn, and K. Ijagbemi, "Project Plan & Product Specifications" Oct. 22, 2015.
"Hardkernel," in ODROID. [Online]. Available: Accessed: Oct. 6,
Rosebrock, Adrian. "Determining Object Color with OpenCV - PyImageSearch." PyImageSearch. 14 Apr Web. 9 Nov
Cross, Nigel. "Engineering Design Methods: Strategies for Product Design, 4th Edition." Engineering Design Methods: Strategies for Product Design, 4th Edition. Wiley, n.d. Web. 7 Nov
K, Abid Rahman. "Simple Digit Recognition OCR in OpenCV-Python." OpenCV-Python. N.p., 01 Jan Web. 8 Nov
Hoblit, Frederick. "Gust Response Equations of Motion: Formulation and Solution." Gust Loads on Aircraft: Concepts and Applications (1988): 8 Nov
Lin, Yucong (Author), and Srikanth (Advisor) Saripalli. "Moving Obstacle Avoidance for Unmanned Aerial Vehicles." ASU Digital Repository. Arizona State University, 01 Jan Web. 8 Nov
Rosebrock, Adrian. "Basic Motion Detection and Tracking with Python and OpenCV - PyImageSearch." PyImageSearch. 07 June Web. 9 Nov
Project Introduction Previous Work Current work conclusion 20

21. Are there any questions?
Project Introduction Previous Work Current work conclusion 21

22. Gantt Chart 1 22

23. Gantt Chart 2 23

24. CPU Ethernet Performance
Figure 13. ODROID Comparison (Mbit/sec). 24

25. Component Justification
Table 6. ODROID Board Specification Comparisons. 25